FOXO1 is an essential regulator of pluripotency in human embryonic stem cells

Abstract

Pluripotency of embryonic stem cells (ESCs) is defined by their ability to differentiate into three germ layers and derivative cell types and is established by an interactive network of proteins including OCT4 (also known as POU5F1; ref. 4), NANOG (refs 5, 6), SOX2 (ref. 7) and their binding partners. The forkhead box O (FoxO) transcription factors are evolutionarily conserved regulators of longevity and stress response whose function is inhibited by AKT protein kinase. FoxO proteins are required for the maintenance of somatic and cancer stem cells; however, their function in ESCs is unknown. We show that FOXO1 is essential for the maintenance of human ESC pluripotency, and that an orthologue of FOXO1 (Foxo1) exerts a similar function in mouse ESCs. This function is probably mediated through direct control by FOXO1 of OCT4 and SOX2 gene expression through occupation and activation of their respective promoters. Finally, AKT is not the predominant regulator of FOXO1 in human ESCs. Together these results indicate that FOXO1 is a component of the circuitry of human ESC pluripotency. These findings have critical implications for stem cell biology, development, longevity and reprogramming, with potentially important ramifications for therapy.

Figure 1

FOXO1 is essential for the expression of hESC pluripotency markers. (a) qRT-PCR analysis of expression of FOXO genes in pluripotent undifferentiated hESCs and during mesodermal induction. The expression levels of FOXO3A and FOXO4 are relative to that of FOXO1 in undifferentiated H1 cells under self-renewal conditions. Note, downregulation of FOXO1 and upregulation of FOXO3A during differentiation of hESCs. FOXO6 expression was not detectable in hESCs. EBs, embryoid bodies. (b) FOXO1 expression was analysed by qRT-PCR in parental H1 hESCs and in H1 cells expressing shRNA targeting FOXO1 (H1/FOXO1 shRNA) cultured in the absence or presence of doxycycline (Dox) for 4 days under undifferentiated self-renewal conditions. H1 cells expressing vector control (H1/VC) or scrambled FOXO1 shRNA (H1/scr) were used as controls. (c) qRT-PCR analysis carried out as in b at the indicated times in cells treated with or without doxycycline. Quantification of the target genes was relative to the endogenous ACTB (-actin) transcript levels. Results are mean s.e.m. of three independent experiments, each carried out in triplicate (a-c); *P < 0.05, *P < 0.01, P < 0.001 (b,c). (d) An aliquot of cells from c was subjected to western blot analysis of the indicated proteins; relative intensities of bands are shown below each panel relative to that measured at time 0 in H1 cells (uncropped scanned gels are shown in Supplementary Fig. S12). (e) hESCs were cultured with or without doxycycline for the indicated times and immunostained for surface markers of pluripotency, TRA-1-60 and TRA-1-81, and counterstained with DAPI. Scale bars, 100 μm.

Figure 2

Reversible effect of loss of FOXO1 on pluripotency and differentiation of hESCs. (a) qRT-PCR analysis of pluripotency (left), mesoderm (middle) and endoderm (right) gene expression in H1/FOXO1-shRNA hESCs. Cells were maintained under pluripotency self-renewing conditions and treated with or without doxycycline for up to 4 days, after which cells were washed extensively and maintained in the absence of doxycycline. Quantification of the genes was relative to the endogenous β-actin transcript levels. Error bars indicate s.e.m. of three independent experiments, each carried out in triplicate. *P <0.05, *P <0.01. (b) In H1 cells, ectopic expression of FOXO1 and not FOXO3A induces expression of pluripotency genes. GFP-positive hESCs lentivirally transduced with an empty control vector or a vector containing FOXO1 or FOXO3A were FACS sorted 72 h after transduction and analysed by qRT-PCR for FOXO1 or FOXO3A and pluripotency marker gene expression; results shown are relative to endogenous ACTB and normalized to untransduced H1 cells under self-renewal conditions. Error bars indicate s.e.m. of three independent experiments each, carried out in triplicate; *P < 0.05, *P < 0.01, *P < 0.001. NT, not transduced. NS, not significant. (c) FOXO1-knockdown-mediated loss of pluripotency and induction of differentiation markers in H1/FOXO1-shRNA cells after several passages. Morphology (left) and alkaline phosphatase staining (right) of hESCs cultured in doxycycline for five passages. (d) Expression of surface markers of pluripotency, SSEA4, TRA-1-60 and TRA-1-81, by immunostaining and DAPI counterstaining of hESCs was analysed after five passages in the absence or presence of doxycycline (Dox). Scale bar, 100 μm. One representative of two independent experiments is shown. (e) qRT-PCR analysis of FOXO1 and pluripotency genes (left), mesoderm (middle) and endoderm (right) markers in H1 cells maintained in pluripotency self-renewal conditions with or without doxycycline for five passages. Results shown are relative to the endogenous ACTB. Error bars indicate s.e.m. (n = 3). *P < 0.05, ***P < 0.001.

Figure 3

Foxo1 and Foxo3 regulate pluripotency of mESCs. (a) Alkaline phosphatase staining in Foxo1-, Foxo3- or control-knockdown mESCs. Scale bar, 50 μm. (b,c) qRT-PCR analysis of pluripotency (top) and developmental genes (bottom) in mESCs expressing one of two distinct shRNA targeting Foxo1 (b) or Foxo3 (c). shRNA-mediated knockdown of luciferase was used as a control. The gene expression levels of four lineages, including trophectoderm (Cdx2 and Mash2, also known as Ascl2), endoderm (Gata6 and Foxa2), ectoderm (Cxcl12, Mash1, also known as Ascl1, and Fgf5) and mesoderm (brachyury) were examined after knockdown of Foxo1 or Foxo3. All graphs show mean ± s.e.m. for n = 3; *P < 0.05, **P < 0.01. P values are from comparing Foxo1 shRNA 4 or Foxo3 shRNA 4 with Luc shRNA 1 (b,c). Ectopic expression of a resistant form of the targeted FoxO1-rescued FoxO1-knockdown-mediated phenotype in both hESCs and mESCs. (d) hESCs were transfected with a lentivirus expressing FOXO1 shRNA III (targeting the 3′ untranslated region of the FOXO1 mRNA); GFP-positive cells were FACS sorted 3 days later (>50-60% GFP positive), transduced with lentiviral vector (pLEIGW-FOXO1) that contains only the FOXO1 coding region and cultured for another 2 days before gene expression analysis. All results are relative to the endogenous ACTB. Error bars indicate s.e.m. (n = 3). *P < 0.05. (e) Endogenous Foxo1 was targeted by Foxo1 shRNA 4 in mESCs and rescued by re-expressing the shRNA-resistant Foxo1-m4 construct (EF1-Foxo1-m4). EF1 promoter empty vector (EF1) was used as a control. Gene expression analysis of Foxo1 and pluripotency genes was carried out by qRT-PCR and all results in Foxo1-shRNA 4 cells are expressed relative to the sample with Luc shRNA and EF1 (set as 1). Error bars indicate s.e.m. (n = 3). P < 0.05 for all genes in rescued samples when compared with controls.

Figure 4

FOXO1 knockdown does not impact hESC proliferation or redox status. (a) Left, H1, H1/VC, H1/scr and H1/FOXO1-shRNA cells maintained in pluripotency self-renewing conditions were cultured with (top) or without (bottom) doxycycline (Dox) for 96 h, after which cell proliferation was analysed by BrdU staining. BrdU incorporation versus DNA content were used to detect percentage of cells that are in G1, S or G2/M phase of cell cycle. Right, quantification of results of two independent experiments is shown. (b) qRT-PCR analysis of anti-oxidant enzymes and anti-oxidant response genes in parental H1 and in H1/FOXO1-shRNA cells maintained with or without doxycycline for 96 h. Quantification of the target genes is relative to the endogenous ACTB transcript levels and normalized to untreated H1 cells under self-renewal conditions. Results are shown as mean ± s.e.m. of triplicate experiments; one representative of two independent experiments is shown. (c) Anti-oxidant treatment does not impact the expression of pluripotency genes in hESCs. qRT-PCR analysis of FOXO1 and pluripotency genes in H1, H1/scr and H1/FOXO1-shRNA cells maintained under pluripotency conditions in the presence or absence of NAC (100 μM) and treated or not with doxycycline for 96 h. Results shown are relative to the endogenous ACTB and normalized to untreated H1/scr or H1 cells under self-renewal conditions. n = 2.

Figure 5

FOXO1 activates the expression of OCT4 and SOX2 pluripotency genes in hESCs by binding directly to their regulatory regions. (a) Sequence alignment of human OCT4 and SOX2 regulatory regions containing putative FoxO-binding sites. (b) Endogenous FOXO occupation of sites shown within arrows was analysed by ChIP in H1 cells. DNA co-immunoprecipitated with either anti-FOXO1 (H-128), anti-FOXO3A (N-15) or a control pre-immune immunoglobulin was amplified by qPCR. Binding of FOXO1 to p27^KIP1 promoter and binding of FOXO1 to a conserved upstream region with no FOXO-binding sequences (in 2.7 kb of human OCT4 (NEG Seq O) and in -4 kb of human SOX2 (NEG - Seq S)) were used, respectively, as positive and negative controls. Results of three independent experiments (mean ± s.e.m.) are shown as relative fold enrichment when compared with control immunoglobulin after normalization to the values obtained from the input samples. (c) ChIP analysis of FOXO1 in undifferentiated H1/FOXO1-shRNA hESCs treated or not with doxycycline (Dox) for 3 days. Results of three independent experiments (mean ± s.e.m.) are shown. (d) pcDNA3, pcDNA3-FOXO1 or pcDNA-FOXO3A were co-transfected into HEK293T cells with a pGL3 containing the O1 or O2 site of human OCT4, or the S1, S2 or S3 site of SOX2 or their mutants driving the luciferase gene; luciferase activity was measured 48 h later. Results are from three independent experiments (mean ± s.e.m.). (e) Luciferase activity measured 48 h after co-transfection of hESCs (H1) with a lentiviral vector control or containing FOXO1 or FOXO3A and a human OCT4 reporter plasmid containing a wild-type or mutant O2 sequence. Results are the mean of two independent experiments. (f) H1 hESCs were maintained continuously (Contin.) in bFGF for 72 h (lane 1) or starved overnight and stimulated or not with bFGF (40 ng ml⁻¹ for 10 min) and/or PI(3)K inhibitor LY294002 (10 μM) for 45 min before stimulation, and preparing the whole-cell lysate (lanes 2-5). One representative immunoblot of two experiments is shown. (g) H1 cells maintained in undifferentiated conditions and treated or not with LY294002 (10 μM) for 2 h before double immunostaining of FOXO1 and phospho-AKT (Ser 473) and counterstaining with DAPI. One representative of three independent experiments is shown (uncropped scanned gels are shown in Supplementary Fig. S12).